Online Time-Resolved Reconstruction Method for Acoustic Tomography System
Online Time-Resolved Reconstruction Method for Acoustic Tomography System
复制标题
DOI：
10.1109/tim.2019.2947949
复制
发表时间：
2020-07-01
影响因子：
5.6
通讯作者：
Jia, Jiabin
中科院分区：
文献类型：
--
作者：
Bao, Yong;Jia, Jiabin
声学层析成像技术可以提供精确的定量重建覆盖的温度分布，设备成本低。对于实时温度场监测的应用，时间分辨率和重建速度都具有重要意义。在这篇文章中，我们开发了一种新的在线时间分辨重建（OTRR）方法，它可以提高时间分辨率，捕捉动态变化，并加快在线实时监测的层析重建过程。首先，利用时间信息的冗余性，基于自适应自攻击（AR）模型设计时间正则化，以减少每帧所需的飞行时间（TOF）数据量。采用滑动重叠窗进一步提高重建精度。其次，递归重建过程在每个数据段上执行滑动迭代。对于每一帧的重建，在线计算是非迭代的。数值模拟和实验室规模的实验进行了验证所提出的OTRR方法。重建图像与基于卡尔曼滤波的OTRR方法进行了比较。结果表明，我们的方法可以提高时间分辨率和计算时间，并产生可接受的结果。
Acoustic tomography can deliver accurate quantitative reconstruction of the covered temperature distribution with low equipment cost. For the application of real-time temperature field monitoring, both the temporal resolution and reconstruction speed are of great significance. In this article, we developed a novel online time-resolved reconstruction (OTRR) method, which can improve temporal resolution to capture dynamic changes and accelerate the tomographic reconstruction process for online real-time monitoring. First, by exploiting the redundancy of the temporal information, a temporal regularization is designed based on adaptive auto aggressive (AR) model to reduce the required amount of time of flight (TOF) data per frame. A sliding overlapping window is applied to further improve the reconstruction accuracy. Second, recursive reconstruction process performs a sliding iteration over each data segment. For the reconstruction of each frame, the online computation is noniterative. Numerical simulation and lab-scale experiment are performed to validate the proposed OTRR method. The reconstruction images are compared with the OTRR methods based on the Kalman filter. The results show that our method can improve the temporal resolution and computational time and produce acceptable results.